知识卡片
马尔可夫状态须足以描述未来
受控过程的Markov条件是给定当前状态S_t和行动A_t后,下一状态与更早历史无关。它不是单说St与历史给定S{t−1}独立;状态若漏掉影响转移的记忆,必须扩展状态或使用部分可观测模型。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
current state is independent of the full history of states given the last state. Contrast this with the causal Markov property of causal DAGs: a node in the DAG is independent of indirect “ancestor” causes given its direct causal parents. We can see that when we view the MDP as a causal DAG, this Markovian assumption is equivalent to the causal Markov property. That means
参考来源
- 位置:《Causal AI》第12章《因果决策与强化学习》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-12.xhtml)
- 结论依据:摘录中关于马尔可夫状态须足以描述未来的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:current state is independent of the full history of states given the last state. Contrast this with the causal Markov property of causal DAGs: a node in the DAG is independent of indirect “ancestor” causes given its direct causal parents. We can see that when we view the MDP as a causal DAG, this Markovian assumption is equivalent to the causal Markov property. That means